robbond
359ccc4ba9
prompt: remove semantic-only mutation conflict
2026-08-11 13:22:29 +01:00
robbond
6adcd817e1
reasoning: require structural progress for supported meaning
2026-08-11 12:32:03 +01:00
robbond
4998de54d2
tooling: enforce no-retry live experiment harness
2026-08-11 11:32:19 +01:00
robbond
0348921542
experiment: add rejected proposal diagnostics to failure path
...
Adds rejectedProposalSnapshot to orchestrator diagnostics for
proposal_compatibility rejections — exposing answerMeaning (userSupportedMeaning,
possibleInference), addedNodes structural fields, addedEdges structural fields,
updatedNodes summaries, and resolvedUnknownNodeIds. Diagnostic evidence only;
does not alter validation, mutation, or error messages. Stage-gated to
proposal_compatibility only.
2026-08-11 08:33:36 +01:00
robbond
2e20d30890
correction: delegate hasNodeLevelUserSupport to rawAnswerSupportsUnclassifiedMeaning
...
v0.15 duplicated the overlap helper logic in hasNodeLevelUserSupport with
reversed argument orientation (unknownText as source, userSupportedMeaning
as candidate) compared to rawAnswerSupportsUnclassifiedMeaning (USM as
source, unknownText as candidate). This produced different accept/reject
outcomes when the two texts have very different token counts.
The fix replaces the independent reconstruction with a single call to the
canonical helper, ensuring node-level and answer-meaning alignment use
identical semantics. Four boundary regression tests verify:
- Boundary A: overlap ratio < 0.4 but >= 3 shared tokens → accept (token rule)
- Boundary B: short candidate / long source accepted via structural linkage
- Boundary B control: unrelated unknown rejected with no structural edge
- Boundary C: ratio exactly at 0.4 threshold accepts via ratio rule
2026-08-10 20:09:39 +01:00
robbond
9e869cbcb8
reasoning: admit verified user-supported unknowns without provenance edges
2026-08-10 19:27:54 +01:00
robbond
0d15dd1f42
Revert "reasoning: require corroboration for conjunction compoundness"
...
This reverts commit 60048a5636
2026-08-10 15:07:30 +01:00
robbond
60048a5636
reasoning: require corroboration for conjunction compoundness
2026-08-10 14:53:57 +01:00
robbond
4c5666dfb3
reasoning: suppress explanation question without relationship structure
2026-08-10 10:17:35 +01:00
robbond
69efc5d1b9
reasoning: ground unclassified answers without category expansion
2026-08-10 08:20:00 +01:00
robbond
7e4c506614
reasoning: prevent unsupported comparison decomposition
2026-08-10 07:44:40 +01:00
robbond
46503b4507
reasoning: normalise graph relationship contract at proposal boundary
2026-08-10 06:16:47 +01:00
robbond
19a42ca7f7
experiment: validate grounded unclassified answer live
2026-08-09 20:03:30 +01:00
robbond
4e4d0fa732
reasoning: stop answer fidelity guard blocking valid unclassified answers
2026-08-09 19:55:35 +01:00
robbond
f861e2cac0
reasoning: preserve evidence versus clarification distinction
2026-08-09 16:20:06 +01:00
robbond
e884b02e7c
experiment: probe user-owned ambiguity boundary
2026-08-09 15:21:02 +01:00
robbond
11882bfaae
experiment: probe evidence versus clarification boundary
2026-08-09 15:11:54 +01:00
robbond
85ee4bed30
experiment: probe explicit hard constraint semantic fidelity
2026-08-09 14:13:48 +01:00
robbond
c40d8c6d49
test: fix canonical live experiment harness import
2026-08-09 12:29:28 +01:00
robbond
e6f784261b
test: establish canonical live reasoning experiment harness
2026-08-09 11:56:10 +01:00
robbond
4aa1492c8d
refine raw-answer boundary for answer meaning
2026-08-09 11:31:27 +01:00
robbond
3e78d57aca
refine answer meaning derivation for negation and qualification
2026-08-09 09:48:08 +01:00
robbond
7965375aff
refine answerMeaning contract around user-supported meaning
2026-08-09 08:50:37 +01:00
robbond
36faf70a08
refine answerMeaning support category normalisation
2026-08-09 08:06:38 +01:00
robbond
0d7ad5775c
reasoning: add proposal-level answer meaning guard
...
- Add answerMeaning schema with supportCategory and resolutionGuidance enums
- Add pre-mutation guard that validates proposal alignment with answerMeaning
- Update prompt builder to instruct the model on answerMeaning contract
- Add tests for schema, guard logic, parsing defaults, and regression cases A-D
2026-08-09 07:07:21 +01:00
robbond
fcb7218407
experiment: separate stated clarification meaning from inference
2026-08-08 08:57:59 +01:00
robbond
3af623a7d5
experiment: test resolution from preserved answer meaning
2026-08-08 08:42:28 +01:00
robbond
22325d5ac3
experiment: separate answer meaning from resolution
2026-08-08 08:23:19 +01:00
robbond
8c12931b43
experiment: test clarification uncertainty preservation
2026-08-08 07:53:30 +01:00
robbond
67f2b084a5
experiment: test clarification broadening with weak answers
2026-08-08 07:43:19 +01:00
robbond
1d1dceefa3
experiment: test consequence of clarification target broadening
2026-08-08 07:29:36 +01:00
robbond
4302e3c435
experiment: test clarification target specificity
2026-08-08 07:16:55 +01:00
robbond
86c04d0fd8
experiment: test end-to-end clarification chain
2026-08-08 06:54:08 +01:00
robbond
5d1cba80cd
experiment: test clarification answer resolution
2026-08-08 06:37:32 +01:00
robbond
8b1d69279f
experiment: test clarification question wording
2026-08-08 06:26:17 +01:00
robbond
c8ead0f690
experiment: test clarification null stability
2026-08-08 06:13:43 +01:00
robbond
6944f358a9
experiment: identify clarification target
2026-08-07 20:03:27 +01:00
robbond
ee1391cd00
experiment: distinguish clarification from evidence needs
2026-08-07 19:28:45 +01:00
robbond
dd3b8e505d
experiment: test structured evidence to consequence reasoning
2026-08-07 19:02:37 +01:00
robbond
cd9328ef8e
experiment: test consequence from explicit evidence needs
2026-08-07 18:32:47 +01:00
robbond
10e87d0d44
experiment: test evidence needs across competing hypotheses
2026-08-07 18:14:03 +01:00
robbond
5daee5e911
experiment: test consequence of interpretation disagreement
2026-08-07 18:01:47 +01:00
robbond
b9f737a293
experiment: test semantic interpretation disagreement
2026-08-07 17:30:10 +01:00
robbond
fb5368ec5f
experiment: test semantic grounding stability
2026-08-07 17:13:41 +01:00
robbond
118c5a789f
experiment: test semantic grounding of interpretations
2026-08-07 16:32:56 +01:00
robbond
0f7457d9c1
experiment: separate source support from interpretation additions
2026-08-07 16:12:29 +01:00
robbond
c2debad66d
experiment: test interpretation lineage to user source
2026-08-07 15:58:39 +01:00
robbond
3537aa1b7a
experiment: test deterministic user source identity
2026-08-07 15:24:47 +01:00
robbond
d0b9be5fe6
experiment: separate stated meaning from model inference
2026-08-07 12:33:49 +01:00
robbond
6af9418eeb
experiment: test grounding of decision relevance
2026-08-07 12:00:10 +01:00
robbond